Polymer Diffusion Is Fastest at Intermediate Levels of Cylindrical

Nov 6, 2018 - The length of the pore is larger than the average polymer chain end-to-end distance, Ree, to prevent chains from interacting with themse...
0 downloads 0 Views 2MB Size
Download PDF
Article Cite This: Macromolecules XXXX, XXX, XXX−XXX

pubs.acs.org/Macromolecules

Polymer Diffusion Is Fastest at Intermediate Levels of Cylindrical Confinement James F. Pressly,† Robert A. Riggleman,*,‡ and Karen I. Winey*,†,‡ †

Department of Materials Science & Engineering, ‡Department of Chemical & Biomolecular Engineering, University of Pennsylvania, Philadelphia, Pennsylvania 19104, United States

Downloaded via LUND UNIV on November 27, 2018 at 21:01:00 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S Supporting Information *

ABSTRACT: The behavior of polymer melts under cylindrical confinement was investigated using molecular dynamics simulations. A range of polymer chains, from unentangled to highly entangled, were confined in cylindrical pores with radii ranging from much smaller to much larger than the polymer size. These simulations were used to measure polymer chain conformation, entanglement density, and center-of-mass diffusion. The conformational anisotropy is well-described by a confined random walk model, although excluded volume effects cause slight differences in the radius of gyration. The number of entanglements per chain in confinement is accurately described using a simple volume fraction model consisting of a zero-entanglement region near the pore wall and a bulklike entanglement region in the pore center. The size of the depletion region near the wall is chain length dependent. Finally, the diffusion along the pore axis exhibits nonmonotonic behavior with the pore radius. As the pore radius decreases, the diffusion coefficient, D, initially increases due to increasing chain disentanglement, though for small pores D eventually decreases as a result of confinement-induced chain segregation.

■

confinement, though fewer studies have examined this area.19 Among the few studies examining cylindrically confined polymer behavior are theoretical models of polymer melts that predict a critical pore size below which polymer chains segregate due to an enhancement of the correlation hole effect.22,25 This predicted segregation is similar to what is observed in ultrathin, 2D polymer melts.5,8,16,17 In contrast to studies of confined chain conformation, there are relatively few works addressing chain-scale dynamics under confinement. There are several studies, both simulations and experiments, that focus on measuring segmental dynamics and changes in the glass transition temperature, though largely in thin film confinement.9,18,30−36 Despite the strong connection between chain conformation and diffusion in polymer melts, few studies have attempted to connect theories and observations of chain conformation in confinement to diffusive behavior in confinement, especially under cylindrical confinement. A notable exception to this generalization is Tung et al., who used experiments and molecular dynamics to examine diffusion of well-entangled chains under cylindrical confinement.27 In this study, we focus on polymer behavior in cylindrical confinement and relate changes in polymer conformation, including chain segregation due to the correlation hole effect,

INTRODUCTION Understanding polymer behavior in confinement has been a focus of polymer physics research for decades.1,2 While it is widely accepted that polymer behavior changes under confinement, the exact nature of the changes and the scaling laws that describe confined polymer behavior are still open for debate. Simulations and theoretical models have been used extensively to understand polymer conformations and dynamics, from local, segmental relaxation to chain diffusion, in a variety of confinement conditions, including thin films,3−21 and cylindrical pores.19,22−28 These simple and uniform geometries are ideal for developing theories that capture the fundamental behavior of confined polymers. Additionally, by comparison of these geometries the role of confining dimensionality is revealed, namely one confining dimension for thin films and two for cylinders. For instance, recent work has shown that while the change in polymer end-to-end distance is similar for polymers confined to thin films and cylinders, the entanglement density exhibits a much greater decrease in cylinders.19 The majority of studies of confined polymers focus on chain conformation in thin films. Studies have shown that polymer conformation perpendicular to the confining surface decreases dramatically while only increasing slightly parallel to the surface.10,11,19 Experiments and simulations have also shown an increase in entanglement molecular weight as film thickness decreases.13,19−21,29 These changes in polymer conformation have been shown to be even stronger in cylindrical © XXXX American Chemical Society

Received: August 13, 2018 Revised: November 6, 2018

A

DOI: 10.1021/acs.macromol.8b01728 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules

0.45 σ less than the nominal radii. The length of the pore is larger than the average polymer chain end-to-end distance, Ree, to prevent chains from interacting with themselves across periodic boundaries. We studied bulk systems with polymer chains of length N = 25, 50, 100, 200, 350, and 500, where N is the number of monomers. Table 1 summarizes the chain lengths studied for each pore size; detailed information can be found in Table S1.

to changes in polymer diffusivity over a range of chain lengths and pore sizes. We first discuss our observations of chain segregation for small pores and compare them to theoretical predictions, showing excellent agreement. Then we express the confined chain conformation using the relative shape anisotropy parameter, κ2, as a function of the confinement parameter, δ. We find that κ2(δ) varies nonmonotonically and that the same behavior occurs whether or not the polymer excluded volume is considered. Interestingly, we demonstrate a nonmonotonic change in diffusion coefficient, D, as a function of cylindrical pore radius and conclude that this is a result of the competing effects of chain disentanglement and chain segregation. Finally, we propose a simple volume fraction model to determine the average number of entanglements per confined polymer chain. Throughout this work, we will demonstrate how polymer conformation and dynamics are strongly linked and how a solid understanding of both is necessary to fully understand confined polymer behavior.

Table 1. Chain Lengths Studied for Each Nominal and Effective Pore Radius nominal radius, r (σ) effective radius, reff (σ) 2.5 3.5 5

■

SIMULATION METHODS The coarse-grained molecular dynamics (MD) simulations were performed using the Kremer−Grest model37 with interactions between nonbonded monomers governed by the repulsive portion of the Lennard-Jones (LJ) potential. The units are normalized to the potential strength, ϵ, monomer size, σ, and time, τ = σ(m/ϵ)1/2, where m is the monomer mass. All simulations were run using the LAMMPS MD simulation package38,39 with the velocity-Verlet algorithm under an NVT ensemble with a monomer density of ϕ = 0.85 σ−3. Confinement was imposed using a pore wall made of discrete immobile particles with the same size and repulsive LJ interaction potential as the polymer monomers. A detailed explanation of the pore wall generation and subsequent polymer equilibration can be found in our previous work.19 A representative image of the confined polymer system is shown in Figure 1.

7 10

14 20

2.05 2.06 3.05 3.07 4.58 4.57 6.57 9.59 9.58 9.57 13.57 19.56

chain length, N 25, 50, 100 200 25, 50, 100 200 25, 100 50, 200, 350, 500 25, 50, 100, 200, 350, 500 25, 50 100 200 25, 50, 100, 200 100, 200

After equilibration, MD simulations were run using a Langevin thermostat, T = 1, with a time step of 0.006 τ until the diffusive regime of the mean-squared displacement (MSD, ⟨r2⟩) was observed and a diffusion coefficient could be calculated. The MSD was calculated from the chain center of mass using a moving time origin. In the confined systems, MSD was calculated along the pore axis (z direction), while in the bulk systems MSD was calculated independently along each axis (x, y, and z) and the results were averaged. To prevent polymer drift from interfering with the MSD calculation, the center of mass of the system was calculated for each frame and shifted to the origin. The diffusion coefficient, D, was calculated from the derivative of the diffusive region of the MSD such that D = (1/2)(d⟨r2⟩/dt). Plots of D = (1/2)(d⟨r2⟩/dt) as a function of t can be found in Figure S3. To study the entanglement properties of our systems we use the Z1 algorithm,40−43 which works using contour-length minimization. The entanglement properties of multiple frames from the simulations were analyzed for each system to generate a sufficiently large set of polymer conformations; the time between successive conformations was on the order of the time for the system to reach the diffusive regime. The average number of entanglements per chain, ⟨Z⟩, was determined using the S-kink method. Measurements of chain radius of gyration, Rg, were performed in a similar manner: analyzing multiple frames, separated by sufficient time, to generate a sufficient number of conformations. In addition to calculating Rg of the polymer, Rg,RW was also calculated for polymer conformations generated from a random-walk (RW) model. In this model, random-walk polymer chain conformations with N steps of length 1.3 were generated in confinement and in the bulk. The step size was chosen to match the bulk Rg from the random-walk model to the MD simulations. In the RW simulations, confinement was defined by smooth cylinders with the same radii used in the MD systems. All chain lengths were simulated in all pore

Figure 1. A representative image showing a confined polymer simulation system (r = 5, N = 350). The pore wall (gray beads) is partially cut away to expose the polymer. Each color represents a distinct polymer chain.

Cylindrical pores were generated in multiple sizes with nominal radii r = 2.5, 3.5, 5, 7, 10, 14, and 20 σ. The effective radius of the pore is defined by the accessible volume, vacc, where reff = (vacc/πL)1/2, in which L is the length of the pore.19,27 The accessible volume is determined by randomly generating points within the simulation box and determining if the point is accessible (inside the pore) or inaccessible (in the pore wall or within a minimum distance of a wall bead). A detailed explanation of the accessible volume calculation can be found in our previous work.19 Because of excluded volume effects, the effective pore radii, reff, are approximately 0.40− B

DOI: 10.1021/acs.macromol.8b01728 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules sizes. The first end of the polymer chain was randomly placed within the pore cross section. New bead positions were rejected if they fell outside the confining cylinder. Randomwalk simulations were also performed with no confining cylinder for comparison with the bulk MD simulations. Because it is a simple random walk, without excluded volume, we can distinguish the effect of excluded volume on polymer conformation in confinement.

■

RESULTS AND DISCUSSION Degree of Confinement and Chain Segregation. We employ the confinement parameter, δ, described by Lee et al.,25 δ = Na2/πreff2, where N is the number of segments in the chain, a is the statistical segment size (based on measurements of the bulk end-to-end distance for a 1D weakly self-avoiding walk, Ree2 = 2Na2), and reff is the radius of the confining pore. A larger δ indicates a greater degree of confinement. Lee et al.25 define several regimes that describe changes in the polymer behavior as confinement increases, including a transition at δ = 1 from weak confinement, where conformational properties are dependent on the shape of the pore cross section, to strong confinement, where these properties are sensitive to only the cross-sectional area. Additionally, the authors define a transition at δ = (ϕreffa2)2 from dispersed to segregated chains. Figure 2a plots δ as a function of reff, demonstrating the range of confinement regimes covered by our simulations. In the segregated chain regime, δ > (ϕRca2)2, the polymers exhibit an extreme version of the correlation-hole effect,44 the phenomenon where there is a reduced concentration of monomers from neighboring chains within the volume occupied by a polymer chain. In this extreme case, the correlation hole is so large that identical polymer chains separate into distinct domains. This is illustrated in Figure 2b, which plots the pair correlation function between chain centers of mass along the pore axis for the smallest pore size. The presence of strong, long-range correlations indicates the polymer chains are segregated. Visualizations of the polymer chains inside the pores clearly show a dispersed chain system (Figure 2c) and a segregated chain system (Figure 2d). In Figure 2d we also note that while there are strong structural correlations on the chain scale, there is no monomer-level ordering. Chain Conformation in Confinement. The average radii of gyration in the unconfined (axial) and confined (radial), directions (Rg,z and Rg,xy, respectively) as a function of the chain length for both the MD and random-walk model simulations are plotted in Figure 3. The average onedimensional radii of gyration in the bulk, R g,bulk1D = R g,bulk / 3 , are also included. In both the MD simulations and the RW model, Rg,xy is always less than Rg,bulk 1D, with their differences increasing with increasing chain length and decreasing pore radius. Additionally, with increasing N, Rg,xy asymptotically approaches a value of reff/2, the radius of gyration of the cylindrical pore cross section. This indicates that the larger chains fully explore all areas of the pore cross section rather than being confined to a small region. In the axial direction we observe very different behavior between the simulations and the model calculations. In MD, Rg,z is greater than Rg,bulk 1D, and similar to the radial direction, their differences are greater for larger chain lengths and narrower pores. In the RW model, Rg,z is equivalent to Rg,bulk 1D

Figure 2. (a) The confinement parameter, δ = Na2/πreff2, as a function of the effective pore radius, reff. The colored domains indicate different confinement regimes. The systems examined span both the weak to strong confinement transition and the transition from dispersed chains (filled markers) to segregated chains (open markers). (b) Pair distribution functions between the polymer chain centers of mass along the z-axis for N = 25, 100 and 200 confined to a pore of r = 2.5 σ and reff = 2.05 σ. The distribution is normalized such that g(r) = 1 indicates a random distribution of chain centers of mass along the z-axis (i.e., dispersed chains). As the chain length increases, the distribution of polymers becomes increasingly correlated along the pore axis, indicating increased segregation of polymers into individual domains. (c, d) Visualizations of polymer chains in confinement for (c) a dispersed chain system (N = 200, r = 7 σ, δ = 1.06) and (d) a segregated chain system (N = 200, r = 2.5 σ, δ = 10.81).

at all N and r values investigated. The r independence of Rg,z, the axial conformation, is due to the absence of excluded volume interactions in the model that are present in the MD simulations. To further examine the anisotropy of confined chain conformation, we calculated the shape anisotropy parameter, κ2, for each chain. The shape anisotropy parameter is defined as κ2 =

4 4 4 3 λ1 + λ 2 + λ3 1 − 2 2 2 2 2 (λ1 + λ 2 + λ3 ) 2

(1)

where λ1 , λ2 , and are the principal moments of the radius of gyration tensor, and their sum is equal to Rg2.45 A κ2 value of 0 indicates a spherically symmetric distribution of monomers, while a value of 1 indicates all monomers lie on a line. The 2

C

2

λ32

DOI: 10.1021/acs.macromol.8b01728 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules

Figure 4. Average relative shape anisotropy, ⟨κ2⟩, of polymer chains for both the MD simulations (closed markers) and the RW model (open markers) plotted against the confinement parameter, δ = Na2/ πreff2. The lack of excluded volume in the RW model causes the simulation and model results to disagree for δ > 4.

and lowering the average relative shape anisotropy. Eventually, as confinement further increases and the system crosses into the strongly confined regime, the principal moment perpendicular to the pore axis becomes less than the other principal moments. At this point the chain becomes primarily oriented along the pore axis, and the relative shape anisotropy begins to increase. In thin films the dip in ⟨κ2⟩ should be much less than in cylinders because there is only one confining dimension. Polymer Diffusion in Confinement. The diffusion coefficients of the confined polymers, normalized to the bulk polymer diffusion coefficient, are plotted as a function of the confinement parameter, δ (Figure 5a). Unlike the shape anisotropy data, the normalized diffusion coefficients, Dnorm, do not collapse to a single curve when plotted as a function of the confinement parameter. This is contrary to polymer diffusion experiments in polymer nanocomposites, where polymers are confined between nanoparticles and the normalized diffusion coefficient is well-described by the degree of confinement as defined by the average interparticle distance.46−48 When the data are replotted as a function of the pore radius (Figure 5b), the diffusive behavior under confinement is nonmonotonic with reff. When polymers are confined to large pores their diffusion coefficient, Dconf, is similar to the bulk diffusion coefficient. As the pore radius decreases (increasing confinement), Dconf increases, and the relative increase is greater for longer chain lengths. For r ≈ 7 σ, Dconf is about 3 times larger than Dbulk for N = 500, while the increase is about 50% for N = 200, and Dconf is nearly unchanged for N = 25 or 50. Dconf continues to increase until the pore radius decreases to about 5 σ, below which Dconf decreases sharply to less than Dbulk. The relative decrease in Dconf is greater for longer chains. For r ≈ 2.5 σ, Dconf is ∼25% of Dbulk for N = 200, while Dconf is about ∼75% of Dbulk for N = 25. The nonmonotonic behavior of Dnorm suggests (at least) two competing factors are affecting confined polymer diffusion. On the basis of the negligible increase in Dnorm for the unentangled systems (N = 25 and 50) and previous studies showing decreased entanglement density in confinement,15,19,20 we conclude that the increase in Dnorm is due to chain disentanglement, as detailed in the next section. The decrease in Dnorm when reff < 5 σ is attributed to chains segregating or beginning to segregate. As shown in Figure 2, chain segregation occurs when the pore is small, such that for

Figure 3. Average Rg values in the confined (radial, shown as filled markers) and unconfined (axial, shown as open markers) dimensions, calculated using (a) MD simulations and (b) random-walk model. Note the difference in the unconfined Rg behavior between the two methods.

results of these calculations are plotted in Figure 4 as a function of δ, the confinement parameter. The ⟨κ2⟩ behavior for the MD simulations and RW model are the same for δ < 4; above this they begin to differ due to the lack of excluded volume effects present in the RW model. As expected, relative shape anisotropy indicates bulklike chain conformations when δ is near zero and more anisotropic conformations for large values of δ. However, the behavior of ⟨κ2⟩ is nonmonotonic with increasing confinement, initially decreasing to a minimum of ∼0.3 near δ = 1, the transition from weak to strong confinement, before sharply increasing. While this is an unexpected result, we explain by considering the orientation of the polymer chains. The primary orientation of a polymer chain can be described using the vector associated with the largest principal moment. In a system where δ is small (unconfined systems), polymer chains are free to orient in any direction. As confinement increases, chains oriented perpendicular to the pore axis are the first to feel the effects of confinement. This causes a slight decrease in that component of the principal moment, causing the chain to adopt a more spherically symmetric conformation D

DOI: 10.1021/acs.macromol.8b01728 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules

which do not benefit from the chain disentanglement that increases the diffusivity of the entangled chains. Instead, in the N = 25 and 50 systems, Dnorm ∼ 1 in all but the smallest pore size. This leads us to believe that any wall friction that might be present in the system has a negligible effect on D compared to chain disentanglement and segregation. With regards to the effect of chain conformation on chain diffusion, we conclude that the average relative shape anisotropy of the polymer chains has a significantly smaller (if any) effect on the diffusion coefficient. While it is certainly possible that for individual chains κ2 affects mobility, given that ⟨κ2⟩ collapses as a function of δ for all chain lengths but the diffusivity does not, the overall diffusive behavior is better explained by changes in the entanglement network and chain segregation. Entanglement Network in Confinement. Understanding the behavior of entanglements in confinement is critical to explaining the diffusive behavior of confined polymer chains shown in Figure 5b. From simulations of free-standing and rigidly confined thin films, we know the entanglement density is reduced near an interface and gradually returns to the bulk density as the distance from the interface increases.15,20 This is the result of having a greater concentration of monomers from the same chain near the pore wall, making entanglements between chains less likely. In Figure 6 we observe that the same holds for polymers confined to cylindrical pores. Figure 6a plots the entanglement density (number of entanglements

Figure 5. Normalized diffusion coefficients plotted (a) as a function of confinement, δ, and (b) as a function of effective pore size. Filled and open markers indicate dispersed and segregated systems, respectively. For reff = 6.57 and 9.59 σ the data points for N = 25 and 50 overlap.

diffusion to occur chains would have to adopt highly stretched and unfavorable conformations. Thus, segregated chains must overcome a very high entropic barrier to diffuse past one another, which dramatically slows chain diffusion along the pore. It should be noted that while the N = 25, r = 2.5 σ system does not show signs of segregation, its diffusion coefficient is still less than bulk. We believe this is because the changes in dynamics as the system transitions from dispersed to segregated is not an abrupt change but rather more gradual. Figure 2a shows that this system is near the segregation transition which is likely causing the chain diffusion to slow, despite the system still being dispersed. We see a similar slowing of diffusion in the N = 100, r = 3.5 σ system, which is also close to the segregated regime. In addition to energetic and entropic barriers, it is possible that wall friction (resulting from the discrete beaded surface of the pore wall) acts on the system as well, causing the dynamics to slow in narrower pores (with higher surface to volume ratios). If there was a sizable wall friction force, we would expect a gradual slowing of D in the unentangled systems,

Figure 6. (a) Entanglement density as a function of distance from the wall for chain length N = 200 showing a reduction in the entanglement density near the pore wall and a maximum at ∼4 σ. The dashed line represents the bulk entanglement density. (b) Entanglement density as a function of distance from the wall for pore radius r = 7 σ. The entanglement depletion region widens with increasing chain length. E

DOI: 10.1021/acs.macromol.8b01728 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules per σ3) as a function of the distance from the pore wall for chain length N = 200 at selected pore radii. We observe the depletion of entanglements near the pore wall, as well as a small peak in the entanglement density at a distance of ∼4 σ from the pore wall, just after the entanglement depleted region, consistent with previous studies.20 It is also important to note that the width of the depleted region at fixed N is independent of the pore radius. Figure 6b plots entanglement density as a function of distance from the pore wall for three chain lengths confined to a pore r = 7 σ. The width of the depleted region strongly depends on the size of the polymer chain; specifically, longer polymer chains have a larger depleted region. These results are consistent with the idea that decreases in interchain entanglement density near the wall are due to a greater concentration of same chain monomers and thus dependent on N and independent of pore size. Note that the entanglement analysis was only performed for dispersed chains. Using these data, we propose an empirical model to describe the average number of entanglements per chain, Z, as a function of pore radius, reff. Upon simplification of the radial dependence of the entanglement distribution in Figure 6 to a step function, where the entanglement density is zero within a distance, td, of the pore wall and is otherwise bulklike, the average number of entanglements per chain in cylindrical confinement is given by Zconf = Z bulk

(reff − td)2 reff 2

(2)

Figure 7. (a) Entanglements per chain, Z, plotted as a function of effective pore radius, reff. The solid lines represent fits to eq 2, where td is the fitting parameter, and the dashed lines represent the number of entanglements per chain in the bulk. (b) Same data as shown in (a), plotted in a linearized form where the data collapses onto a line of slope −1, consistent with eq 2 (inset). The width of the entanglement depleted region, td, as a function of chain length, N, which varies approximately as N1/2.

where Zconf and Zbulk are the number of entanglements per chain in the confined and bulk systems, respectively, and td is the thickness of the entanglement depleted region. In Figure 7a, Zconf is plotted as a function of reff for r = 3.5− 20 σ and N = 100−500. (Data from segregated chain systems are excluded.) The solid lines are fits of eq 2 to Zconf for each chain length, using td as the sole fitting parameter: td(N = 100) = 0.81 σ, td(N = 200) = 1.04 σ, td(N = 350) = 1.34 σ, and td(N = 500) = 1.51 σ. This simple model of entanglement distributions in cylindrical confinement is in excellent agreement with the number of entanglements per chain. Figure 7b displays the same entanglement data plotted in the linearized form of eq 2: td/reff vs (Zconf/Zbulk)1/2. From the inset in Figure 7b, note that td is consistent with the scaling of N1/2 and td ≈ 0.14Rg,bulk. This finding qualitatively agrees with our observation that the width of the depletion region increases with N as seen in Figure 6b. Overall, it is intriguing that the width of the entanglement depletion region is constant relative to Rg rather than related to the entanglement length. However, in all cases, td is less than the tube diameter (∼1.8 σ in bulk), and it is plausible this trend might saturate as td approaches the tube diameter. The observations of entanglement density in this study are in qualitative agreement with previous work observing a decrease in entanglement density with increasing confinement.15,19,20 We have expanded on findings of these studies by noting that the width of the entanglement depletion region scales with chain size. Relating the entanglement behavior back to the diffusive behavior, reptation theory states that the number of chain entanglements is inversely proportional to the diffusion coefficient, D ∝ Z−1. A qualitatively similar trend occurs in cylindrical confinement. As the average number of entanglements per chain decreases, the diffusion coefficient increases.

Longer chains lose a larger fraction of their entanglements compared to shorter chains and thus exhibit a greater increase in their diffusion coefficient compared to shorter chains (Figure 5b). Our simulations for r > 5 σ support the explanation that the increase in the mobility of cylindrically confined polymers is due to chain disentanglement. Effect of Chain Segregation on Confined Diffusion. When reff < 5 σ a secondary effect is present, namely, chain segregation within the cylindrical pore, that dramatically slows polymer diffusion. Comparing the confinement parameters and associated confinement regimes for each system as shown in Figure 2a to the diffusion data shown in Figure 5b, we observe that all of the systems that show a decrease in Dnorm when the pore radius is small (all r = 2.5 σ systems and N = 100 and 200 for r = 3.5 σ) are within the segregated regime or near the dispersed-to-segregated boundary. For the r = 2.5 σ systems, the decrease in the diffusion coefficient relative to bulk is greater for longer chains which, as shown in Figure 2b, are more strongly segregated. This is consistent with our explanation that the decreased diffusion coefficients are caused by chain segregation. For segregated chains to diffuse past one another along the pore they must elongate to exchange positions with neighboring chains, and this reduces their conformational entropy. To quantify the strength of the entropic barrier, we performed a potential of mean force analysis on the segregated and near segregated systems. Using the equation F

DOI: 10.1021/acs.macromol.8b01728 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules w(z) = −kT ln g (z) = u(z) − s(z)T

entropic barriers, we observe no evidence of caging when examining the mean-squared displacement data (Figure S2), even in the most confined systems.

(3)

where w(z) is the potential of mean force between two polymer chains with centers of mass separated by distance z along the pore axis where z is always taken in the positive direction, g(z) is the chain center of mass distribution function along the pore axis, and u(z) and s(z) are the interaction energy and entropy between two chains separated by distance z. The Boltzmann constant, k, and temperature, T, are both 1. To calculate the entropic barrier, u(z) was calculated between pairs of polymer chains and binned by the distance between the chain centers of mass. This distribution was then subtracted from w(z) to determine −s(z). Figure 8a shows the entropy of two chains with centers of mass separated by distance z along the pore axis for the r =

■

CONCLUSIONS We used molecular dynamics simulations to examine the chain conformations, entanglement densities, and diffusion coefficients of polymer chains under athermal cylindrical confinement. The simulation systems covered a range of chain lengths, N = 25−500, spanning the unentangled and entangled regimes. By using a variety of pore sizes, r = 2.5−20 σ, confinement spanned both weak and strong regimes as well as systems undergoing confinement induced chain segregation. Additionally, a random walk (RW) model was applied over the same range of chain length and pore sizes to isolate the effects of excluded volume and geometric confinement on chain conformations. To examine the confined chain conformations, the shape anisotropy of the average Rg vectors was calculated for each system in MD simulations and confined random walk models. We observed that in the MD simulations Rg,z > Rg,bulk 1D and Rg,xy < Rg,bulk 1D, while in the random-walk model Rg,z = Rg,bulk 1D and Rg,xy < Rg,bulk 1D. Despite the differences in confined Rg behavior in the MD simulations and RW model, the ⟨κ2⟩ behavior was similar for a given degree of confinement, with each system showing a dip in ⟨κ2⟩ at δ = 1. Thus, the shape anisotropy parameter is a robust measure of confined chain conformation, dependent only on the degree of confinement, δ. Calculating the shape anisotropy parameter in additional pore geometries (e.g., square or rectangular cross sections) will enable us to determine the generalizability of this κ2 behavior. The entanglement density as a function of radial position was used to develop a simple two-layer model (eq 2) that captures the number of entanglements per chain as a function of the pore radius and the chain length. Examining entanglement behavior in additional geometries will determine whether it is possible to generalize our two-layer model. The diffusive behavior was examined by normalizing the axial diffusion coefficient to the bulk diffusion coefficient, and it was shown that Dnorm changes nonmonotonically as a function of the pore radius. This is in contrast to our early experimental and simulation work. By extending to more confined systems and chain lengths, this paper demonstrates the competing effects of chain disentanglement increasing diffusivity and chain segregation decreasing diffusivity. Because chains are slowed due to the increased free energy barrier for segregated chains to diffuse past each other, the dramatic decrease in Dnorm for small reff may be particular to cylindrical confinement. In thin film (1D) confinement, this effect of chain segregation might be less pronounced.

Figure 8. (a) Entropy and (b) potential of mean force for two polymer chains separated by a distance z in a pore of radius 2.5 σ. Note the large energetic and entropic barrier as the two chains come into contact with one another. This barrier decreases as chain length decreases and as pore radius increases.

2.5 σ system. For the most confined system (r = 2.5 σ and N = 200) the entropic barrier for two chains to pass is on the order of 18kT. The free energy distribution, shown in Figure 8b, indicates the while the entropic barrier is very high, the free energy barrier for chains to pass is only ∼5kT. Both the free energy and entropic barriers quickly decrease as the chain length decreases and the pore radius increases. For example, the energy barrier decreases from 5kT to 2.5kT for N = 100 and from 5kT to 1kT for r = 3.5 σ (Figure S1). These findings corroborate the conclusions of Lee et al.,25 who predicted nonmonotonic diffusive behavior due to a strong slowing down of the dynamics when approaching the dispersed to segregated transition due to the extra stretching required of the chains to move past one another. Despite the significant energetic and

■

ASSOCIATED CONTENT

* Supporting Information S

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.macromol.8b01728. Figures S1−S3 and Table S1 (PDF)

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail [email protected]. *E-mail [email protected]. G

DOI: 10.1021/acs.macromol.8b01728 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules ORCID

(17) Wittmer, J. P.; Meyer, H.; Johner, A.; Kreer, T.; Baschnagel, J. Algebraic Displacement Correlation in Two-Dimensional Polymer Melts. Phys. Rev. Lett. 2010, 105, 037802. (18) Shavit, A.; Riggleman, R. A. Physical Aging, the Local Dynamics of Glass-Forming Polymers under Nanoscale Confinement. J. Phys. Chem. B 2014, 118, 9096−9103. (19) Sussman, D. M.; Tung, W.-S.; Winey, K. I.; Schweizer, K. S.; Riggleman, R. A. Entanglement Reduction and Anisotropic Chain and Primitive Path Conformations in Polymer Melts under Thin Film and Cylindrical Confinement. Macromolecules 2014, 47, 6462−6472. (20) Sussman, D. M. Spatial Distribution of Entanglements in Thin Free-Standing Films. Phys. Rev. E: Stat. Phys., Plasmas, Fluids, Relat. Interdiscip. Top. 2016, 94, 012503. (21) Lee, N.-K.; Diddens, D.; Meyer, H.; Johner, A. Local Chain Segregation and Entanglements in a Confined Polymer Melt. Phys. Rev. Lett. 2017, 118, 067802. (22) Brochard, F.; de Gennes, P. G. Dynamics of Confined Polymer Chains. J. Chem. Phys. 1977, 67, 52−56. (23) Brochard, F.; de Gennes, P. Conformations de polymères fondus dans des pores très petits. J. Phys., Lett. 1979, 40, 399−401. (24) Brochard-Wyart, F.; Raphael, E. Scaling Theory of Molten Polymers in Small Pores. Macromolecules 1990, 23, 2276−2280. (25) Lee, N.-K.; Farago, J.; Meyer, H.; Wittmer, J. P.; Baschnagel, J.; Obukhov, S. P.; Johner, A. Non-Ideality of Polymer Melts Confined to Nanotubes. EPL (Europhysics Letters) 2011, 93, 48002. (26) Carrillo, J.-M. Y.; Sumpter, B. G. Structure and Dynamics of Confined Flexible and Unentangled Polymer Melts in Highly Adsorbing Cylindrical Pores. J. Chem. Phys. 2014, 141, 074904. (27) Tung, W.-S.; Composto, R. J.; Riggleman, R. A.; Winey, K. I. Local Polymer Dynamics and Diffusion in Cylindrical Nanoconfinement. Macromolecules 2015, 48, 2324−2332. (28) Polson, J. M.; Tremblett, A. F.; McLure, Z. R. N. Free Energy of a Folded Polymer under Cylindrical Confinement. Macromolecules 2017, 50, 9515−9524. (29) Si, L.; Massa, M. V.; Dalnoki-Veress, K.; Brown, H. R.; Jones, R. A. L. Chain Entanglement in Thin Freestanding Polymer Films. Phys. Rev. Lett. 2005, 94, 127801. (30) Frank, B.; Gast, A. P.; Russell, T. P.; Brown, H. R.; Hawker, C. Polymer Mobility in Thin Films. Macromolecules 1996, 29, 6531− 6534. (31) Zheng, X.; Rafailovich, M. H.; Sokolov, J.; Strzhemechny, Y.; Schwarz, S. A.; Sauer, B. B.; Rubinstein, M. Long-Range Effects on Polymer Diffusion Induced by a Bounding Interface. Phys. Rev. Lett. 1997, 79, 241−244. (32) Dalnoki-Veress, K.; Forrest, J. A.; Murray, C.; Gigault, C.; Dutcher, J. R. Molecular Weight Dependence of Reductions in the Glass Transition Temperature of Thin, Freely Standing Polymer Films. Phys. Rev. E: Stat. Phys., Plasmas, Fluids, Relat. Interdiscip. Top. 2001, 63, 031801. (33) Pu, Y.; Rafailovich, M. H.; Sokolov, J.; Gersappe, D.; Peterson, T.; Wu, W.-L.; Schwarz, S. A. Mobility of Polymer Chains Confined at a Free Surface. Phys. Rev. Lett. 2001, 87, 206101. (34) Tsui, O. K. C.; Zhang, H. F. Effects of Chain Ends and Chain Entanglement on the Glass Transition Temperature of Polymer Thin Films. Macromolecules 2001, 34, 9139−9142. (35) Yang, Z.; Peng, D.; Clough, A.; Lam, C.-H.; Tsui, O. K. C. Is the Dynamics of Polystyrene Films Consistent with Their Glass Transition Temperature? Eur. Phys. J.: Spec. Top. 2010, 189, 155−164. (36) Yang, Z.; Fujii, Y.; Lee, F. K.; Lam, C.-H.; Tsui, O. K. C. Glass Transition Dynamics and Surface Layer Mobility in Unentangled Polystyrene Films. Science 2010, 328, 1676−1679. (37) Kremer, K.; Grest, G. S. Dynamics of Entangled Linear Polymer Melts:AMolecular-Dynamics Simulation. J. Chem. Phys. 1990, 92, 5057−5086. (38) Plimpton, S. Fast Parallel Algorithms for Short-Range Molecular Dynamics. J. Comput. Phys. 1995, 117, 1−19. (39) Auhl, R.; Everaers, R.; Grest, G. S.; Kremer, K.; Plimpton, S. J. Equilibration of Long Chain Polymer Melts in Computer Simulations. J. Chem. Phys. 2003, 119, 12718−12728.

James F. Pressly: 0000-0002-9365-9562 Robert A. Riggleman: 0000-0002-5434-4787 Karen I. Winey: 0000-0001-5856-3410 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS J.F.P., K.I.W., and R.A.R. acknowledge support from DOE-BES via DE-SC0016421. Computational resources were made available at the National Institute for Computational Sciences, Texas Advanced Computing Center, and San Diego Supercomputer Center through XSEDE Award TG-DMR150034.

■

REFERENCES

(1) McKenna, G. B. Ten (or More) Years of Dynamics in Confinement: Perspectives for 2010. Eur. Phys. J.: Spec. Top. 2010, 189, 285−302. (2) Lin, C.-C.; Parrish, E.; Composto, R. J. Macromolecule and Particle Dynamics in Confined Media. Macromolecules 2016, 49, 5755−5772. (3) Kumar, S. K.; Vacatello, M.; Yoon, D. Y. Off-Lattice Monte Carlo Simulations of Polymer Melts Confined between Two Plates. J. Chem. Phys. 1988, 89, 5206−5215. (4) Bitsanis, I.; Hadziioannou, G. Molecular Dynamics Simulations of the Structure and Dynamics of Confined Polymer Melts. J. Chem. Phys. 1990, 92, 3827−3847. (5) Carmesin, I.; Kremer, K. Static and Dynamic Properties of TwoDimensional Polymer Melts. J. Phys. (Paris) 1990, 51, 915−932. (6) Kumar, S. K.; Vacatello, M.; Yoon, D. Y. Off-Lattice Monte Carlo Simulations of Polymer Melts Confined between Two Plates. 2. Effects of Chain Length and Plate Separation. Macromolecules 1990, 23, 2189−2197. (7) Pakula, T. Computer Simulation of Polymers in Thin Layers. I. Polymer Melt between Neutral Walls − Static Properties. J. Chem. Phys. 1991, 95, 4685−4690. (8) Müller, M. Chain Conformations and Correlations in Thin Polymer Films: A Monte Carlo Study. J. Chem. Phys. 2002, 116, 9930−9938. (9) Varnik, F.; Baschnagel, J.; Binder, K. Reduction of the Glass Transition Temperature in Polymer Films: A Molecular-Dynamics Study. Phys. Rev. E: Stat. Phys., Plasmas, Fluids, Relat. Interdiscip. Top. 2002, 65, 021507. (10) Cavallo, A.; Müller, M.; Binder, K. Unmixing of Polymer Blends Confined in Ultrathin Films: Crossover between Two-Dimensional and Three-Dimensional Behavior. J. Phys. Chem. B 2005, 109, 6544− 6552. (11) Cavallo, A.; Müller, M.; Wittmer, J. P.; Johner, A.; Binder, K. Single Chain Structure in Thin Polymer Films: Corrections to Flory’s and Silberberg’s Hypotheses. J. Phys.: Condens. Matter 2005, 17, S1697. (12) Romiszowski, P.; Sikorski, A. Dynamics of Polymer Chains in Confined Space. A Computer Simulation Study. Phys. A 2005, 357, 356−363. (13) Meyer, H.; Kreer, T.; Cavallo, A.; Wittmer, J. P.; Baschnagel, J. On the Dynamics and Disentanglement in Thin and Two-Dimensional Polymer Films. Eur. Phys. J.: Spec. Top. 2007, 141, 167−172. (14) Sikorski, A.; Romiszowski, P. Computer Simulations of Polymers in a Confined Environment. J. Phys.: Condens. Matter 2007, 19, 205136. (15) Vladkov, M.; Barrat, J.-L. Local Dynamics and Primitive Path Analysis for a Model Polymer Melt near a Surface. Macromolecules 2007, 40, 3797−3804. (16) Meyer, H.; Kreer, T.; Aichele, M.; Cavallo, A.; Johner, A.; Baschnagel, J.; Wittmer, J. P. Perimeter Length and Form Factor in Two-Dimensional Polymer Melts. Phys. Rev. E 2009, 79, 050802. H

DOI: 10.1021/acs.macromol.8b01728 Macromolecules XXXX, XXX, XXX−XXX

Article

Macromolecules (40) Kröger, M. Shortest Multiple Disconnected Path for the Analysis of Entanglements in Twoand Three-Dimensional Polymeric Systems. Comput. Phys. Commun. 2005, 168, 209−232. (41) Shanbhag, S.; Kröger, M. Primitive Path Networks Generated by Annealing and Geometrical Methods: Insights into Differences. Macromolecules 2007, 40, 2897−2903. (42) Hoy, R. S.; Foteinopoulou, K.; Kröger, M. Topological Analysis of Polymeric Melts: Chain- Length Effects and Fast-Converging Estimators for Entanglement Length. Phys. Rev. E 2009, 80, 031803. (43) Karayiannis, N. C.; Krö ger, M. Combined Molecular Algorithms for the Generation, Equilibration and Topological Analysis of Entangled Polymers: Methodology and Performance. Int. J. Mol. Sci. 2009, 10, 5054−5089. (44) De Gennes, P.-G. Scaling Concepts in Polymer Physics; Cornell University Press: Ithaca, NY, 1979. (45) Theodorou, D. N.; Suter, U. W. Shape of Unperturbed Linear Polymers: Polypropylene. Macromolecules 1985, 18, 1206−1214. (46) Choi, J.; Hore, M. J. A.; Meth, J. S.; Clarke, N.; Winey, K. I.; Composto, R. J. Universal Scaling of Polymer Diffusion in Nanocomposites. ACS Macro Lett. 2013, 2, 485−490. (47) Choi, J.; Hore, M. J. A.; Clarke, N.; Winey, K. I.; Composto, R. J. Nanoparticle Brush Architecture Controls Polymer Diffusion in Nanocomposites. Macromolecules 2014, 47, 2404−2410. (48) Lin, C.-C.; Gam, S.; Meth, J. S.; Clarke, N.; Winey, K. I.; Composto, R. J. Do Attractive Polymer−Nanoparticle Interactions Retard Polymer Diffusion in Nanocomposites? Macromolecules 2013, 46, 4502−4509.

I

DOI: 10.1021/acs.macromol.8b01728 Macromolecules XXXX, XXX, XXX−XXX